The Basic And Applied Research Of Independent Component Analysis In Spectral Analysis

Nowadays it’s a common phenomenon that counterfeit drugs are manufactured andsold, which results in the safety of the drug often cannot be guaranteed, seriouslyendangering the health of the people and disrupting the proper and orderly competition ofthe pharmaceutical market. Drug detection provides necessary technical support for thepharmaceutical supervision and administration, and how to improve the technology andefficiency of the fast identification of counterfeit drugs in the supervision has been a toppriority. Due to various types of fraud, the technology of the fast identification ofcounterfeit drugs is facing serious challenges. Identification of counterfeit drugs isusually via a series of qualitative and quantitative methods such as chemical reaction,thin layer chromatography, liquid chromatography and hyphenated techniques, whoseresults have to be verified by each other to increased Its Credibility. The needing of along time of test, a large number of samples and mostly destruction limits thepopularization of these methods. In recent years, some accurate, rapid and convenientmethods need to be used in routine screening of drugs. With the development ofchemometrics, spectroscopy attracts extensive attention in the world; Near infraredspectra do not have a ideal accuracy because of the serious overlap of the spectral line;with a high testing requirements of environmental temperature and humidity, a longcycle of detection, and a complex pre-treatment of sample, the infrared spectroscopycan’t meet the requirements of fast identification; Raman spectroscopy can reflect theskeleton of compounds vibration and be relatively suitable to the identification ofcounterfeit drugs. Particularly, with a small, lightweight and convenient body, needingsmall amount of sample, without pre-treatment, and non-destructive test, portable Ramanspectrometer have some advantages in this area.Some commercial hypoglycemic tablets, capsules, common drug excipients andtheir corresponding active pharmaceutical ingredients were used as tool samples in thispaper, we establish and improve the method, application and solution of the independent component analysis based on the Raman spectra, thus realizing the quick analysis of aparticular compound (really medicine composition, fake composition) in complex matrix.The process from the collection of sample spectra to obtain analytical results can be akey to complete in this program which has the potential to be a universal screeningtechnology of the counterfeit and inferior drugs. The program based on spectralinformation science proposes the new way of thinking and can update the understandingof―model-free analysis‖in academia; it will play a positive role in promoting the basicresearch study, as well as expand into other areas, such as the food and drug safety.Independent component analysis has been a statistical signal processing methods fornearly30years, it mainly used in blind source separation and feature extraction; since1995, ICA has become a powerful algorithm for the separation from the complex system.The paper includes two aspects:1. Learning from a combination model of the ICAand near-infrared or infrared spectra, we are intend to establish and improve the method,application and solution of the independent component analysis based on the Ramanspectra using Matlab programming, and use it in the fast identification of counterfeitdrugs. On the basis of fast detection method of our group, we are to broaden our thinkingand expect to form a more perfect counterfeit fast detection system.2. In this paper, a fullrange of comparative study is made among three portable Raman instrumentations fromthree different companies (Thermo TruScan handheld Raman, Ocean Optics QE65000-Raman and B&W Tek i-Raman).Part Ⅰ The study of counterfeit fast detection method based on independentcomponent analysis.(1)5kinds of hypoglycemic reference compounds,6kinds ofhypoglycemic drugs, and12kinds of drug excipients were identified using independentcomponent analysis of their Raman spectra.90.15%of the hypoglycemic tablets werepredicted correctly.(2) In this paper, a “model-free analysis‖method based onindependent component analysis and Raman spectroscopy is developed to distinguishcounterfeit drugs. To adapt to the detection of the suspected counterfeit drugs based on portable Raman spectroscopy, some modifications are made in this paper to improve theICA algorithm. The algorithms are exemplified by6kinds (43batches) of hypoglycemicdrugs and12common pharmaceutical excipients. The total sensitivity, specificity andaccuracy reach92.11%,89.36%and96.55%respectively.Part ⅡComparative study of portable Raman instrumentation for pharmaceuticalanalysis. In this paper, a comparative study is made among three portable Ramaninstrumentations from three different companies (Thermo TruScan handheld Raman,Ocean Optics QE65000-Raman and B&W Tek i-Raman), in terms of external properties,hardware performance, software function, and analytical method validation. Thespectrometers were obtained from Thermo (the original Ahura Scientific), Ocean Opticsand B&W Tek Inc. The Raman spectra of six commercial hypoglycemic drugs and theircorresponding reference compounds of active pharmaceutical ingredients (APIs) wereacquired by the three spectrometers under as similar conditions as possible and were thencompared with the―JY-reference spectra‖, which were collected using a Jobin YvonLabRam-1b micro-Raman spectroscopy.